Abstract: In the actual production activities,production plan and equipment's maintenance plan are made by different departments,which can cause a conflict and increase a high cost inevitably.To formulate production scheduling plan and preventive maintenance scheme,joint optimization problem of these two plans in the batch-discrete re-enter processing environment was researched,in which both batch processor and discrete processor were existed.Batch processor could process several jobs simultaneously,but discrete processor could just process jobs one by one and many re-enter process flows were existed.An Adaptive Differential Evolution algorithm based on Information Entropy(ADEIE)was proposed.In this algorithm,the distance between individual solution and the potential optimal solution was regarded as a kind of information.Through the measurement of information,the individual variation level was adjusted for effectively reducing the scope of search space.Compared with the other 4 algorithms,the simulation experimental result showed that the ADEIE was superior to other algorithms in either quality of solution or convergence rate.

Key words: batch processor, discrete processor, preventive maintenance, adaptive differential evolution algorithm, information entropy, production scheduling

摘要： 针对现实生产活动中生产计划和设备维护计划由不同部门制定,易造成方案冲突、计划变动成本高的问题,为统筹生产调度和设备预防性维护计划,研究了半导体制造中批—离散机重入车间调度与设备预防性维护联合优化。该问题中同时存在多台批处理机和离散机,其中批处理机能同时加工多个工件,离散机一次只能加工一个工件,且存在大量的重入工序。提出一种求解该问题的基于信息熵的自适应差分进化算法,该算法将个体与解空间中潜在最优解之间的距离映射为信息,通过对信息的度量调整个体变异的程度,从而有效缩小搜索空间。与其他4种算法进行比较,结果表明,该算法无论是求解质量还是收敛速度均优于其他算法。

关键词: 批处理机, 离散机, 设备维修, 自适应差分进化算法, 信息熵, 生产调度